For the past two and a half years following the launch of a transformational collaboration with Genentech and Roche, a team of 50 people at Recursion has had a singular mission – deliver the world’s first Neuromap, developing the neuronal data and the machine learning models to drive new insights in neurodegenerative diseases.
For those working on the project in its earliest days, the effort began with a heady combination of uncertainty and hope, says team leader Kelly Zalocusky, Senior Director of Computational Biology and site lead of Recursion’s London office.
“We were enthusiastic, but nobody was sure we could make enough neurons, knock out every gene, and get reliable enough signal from our machine learning models to start drug programs and make decisions that might impact human life,” Zalocusky said. “This is a different kind of bar.”
Recursion first developed its biological map-making methods in 2013 using HUVEC (human umbilical vein endothelial cells) cells – creating over 100 billion cells per year for high throughput experiments to become one of the largest HUVEC cell producers in the world. These cells are “workhorse” cells – inexpensive and easy to use to model different disease types. But they’re not relevant in all diseases.
The team expanded those efforts using GI-oncology cells to create GI-oncology specific maps as part of our partnership agreement with Roche and Genentech. That effort was successful and that collaboration had its first program optioned in October 2023.
Still, they knew that building the first Neuromap would be an order of magnitude more challenging. Neuronal cells, which do not divide and proliferate, are not nearly as easy to produce. Recursion collaborated with Genentech to develop and adapt a human-induced pluripotent stem cells (or hiPSC)-derived neuronal model and protocol for the creation of a unique phenomap that can be utilized to uncover novel insights in neuronal biology.
The Recursion team then built specific cell manufacturing technologies that could derive neurons from the hiPSCs at scale – ultimately producing over 1 trillion hiPSC-derived neuronal cells in order to have enough to build a map.
Kelly Zalocusky, Senior Director of Computational Biology and site lead of Recursion’s London office.
“We’re at a time in the history of drug development that it’s just starting to be the case that we can rationally pursue neuroscience,” says Zalocusky. “Oncology is ahead by 20 years. It’s been so hard to get treatments for these patients with neurodegeneration, and everyone has been following the same hypotheses. The map gives us an unbiased view of relationships between the many genes in the Neuromap that exhibit a phenotype. We’re not just looking for our keys under the lamppost.”
The Neuromap is an exciting new way to think about human biology, Zalocusky said, noting also that our partners are seeing biology that is known to neuroscience experts as well as potentially compelling biology that is not known using the map. “It’s important to see these examples that you trust as well as new possible targets,” she says.
Under this collaboration which leverages the Recursion OS and expertise in phenotypic screening as well as Genentech’s expertise in neuronal biology, next steps will be for the joint team to identify potential new novel targets to validate. Even as these avenues are being explored, the team is also pursuing the development of even more robust Neuromaps. This first map is a genetics-only map, with the whole genome knockout and many additional perturbations related to neurodegenerative diseases. Future maps could comprise additional cell types or large compound libraries.
“Working with Roche and Genentech, we have the opportunity to go after big projects,” says Zalocusky, “the largest unmet need in terms of patient population and impact to society – the greatest good for the greatest number of people.”
#ai #maps #neurons #neurological #neuroscience #drugdiscovery #techbio
Author: Brita Belli, Senior Communications Manager at Recursion.
